I have a scenario at the data level there is a n..n relation between fact and dimension...How would it hurt the results if i have the cardinality as 1..n in the Fm even though at the data level it remains n..n
The purpose of defining the cardinality in FM is to tell FM which table should be treated as a Dimension and which one as a Fact. The table on the 1 side will be treated as a Dimension and the one on N side will be treated as a Fact. It does not have anything to do with the number of rows of one table to be joined to number of rows of another table.
That being said, if you have more than one row in the Dimension that will be joined to the same row of the Fact then you have a serious problem to deal with on the data side. The aggregation would be unpredictable.
What if there is no aggregation and its purely dimensional. two dimensions getting joined which have n..n in the data
If you do not use any measures from a fact, then Cognos will automatically fetch distinct values from the query. However, you would not normally fetch data from directly related dimensions. So, to answer your question from a different perspective: all combinations generated by the join strategy will be fetched, but duplicates will be removed from the resultset..
Fact to Fact relationships can be resolved by implementing a junction table query subject between the two tables.